>>It turns out that most of the advantage for FX cameras >comes >>from simply having a 2.25x larger sensor, which collects >2.25x >>more light. > >Your logic is wrong: A given lens transmits a certain and >limited amount of light (= photons) per time unit. If you have >a larger senser and more pixels, you spread the same amount of >light over a larger area, which gives LESS amount of light >(photoms) per pixel, and the signal-to-noise level will drop, >which will also decrease the dynamic range. It's simple >physics.
It is physics and it is simple, yet when your conclusions aren't actually borne out in real world usage, maybe it's time to examine the assumptions:
In the real world you don't shoot "a given lens," you compose a picture. Why is this relevant? Because you'll compose your image with a different lens for an FX sensor than you will for a DX sensor.
Let's for example say you compose a picture with a 35mm lens at f/8 on a DX camera. Your exit pupil is 35mm/8 = 4.35mm. Now compose the same picture on FX, you'll use a 50mm lens at f/8. Your exit pupil is 6.25mm. 1.5x larger diameter = 2.25x larger area = 2.25x more light.
Can you explain how a D800 beats a D7000 by more than a full stop in dynamic range, despite (according to DxO) both cameras having nearly identical pixel-level performance?